LLM4XAI 2026: The 1st Workshop on Generative Models for Explainable AI Narratives

⸻

LLM4XAI 2026 IN SHORT

Generative Models for Explainable AI Narratives

A Half-Day Workshop co-located with the 35th International ACM Conference on Information and Knowledge Management (CIKM 2026)

November 8, 2026 – Rome, Italy

Website: https://hercolelab.github.io/LLM4XAI/index.html

⸻

ABOUT LLM4XAI 2026

The rapid advancement of Large Language Models (LLMs) is opening new opportunities for Explainable Artificial Intelligence (XAI), enabling the generation of natural-language explanations that can make complex AI decisions more accessible to end users. At the same time, the emergence of generative approaches for XAI raises fundamental challenges concerning the faithfulness, reliability, robustness, and alignment of generated explanations with the underlying model behavior.

LLM4XAI aims to bring together researchers and practitioners from XAI, Natural Language Processing, Information Retrieval, Human-Computer Interaction, and related fields to discuss emerging methodologies, challenges, and applications of LLM-based approaches for generating reliable, trustworthy, and human-centered AI explanations.

The workshop will feature original research paper presentations, invited talks, and interactive discussions aimed at fostering collaboration and advancing research on generative models for explainable AI narratives.

TOPICS OF INTEREST

Authors are invited to submit papers that fall into or are related to one or multiple topic areas listed below:

* Explanation Narratives
- From artifacts to narratives: turning attributions, counterfactuals & saliency into natural language
- Multimodal explanations (text, visuals, tables, code)
- Agentic & multi-agent explanation pipelines
- Causal and structured grounding of explanations
- Narrative planning, composition & audience framing
- Retrieval- and tool-augmented explanation generation
* Trustworthy & Interpretable Explanations
- Faithfulness and factual consistency of narratives
- Detecting and mitigating hallucinated explanations
- Mechanistic interpretability & circuit-level analysis
- Aligning narratives with the model's actual reasoning
- Evaluation: automatic metrics, human studies & benchmarks
- Reproducibility and robustness
* Human-Centered & Responsible XAI
- Interactive and conversational explanations
- Personalized, audience-adaptive explanations
- Actionable recourse and human-in-the-loop decisions
- High-stakes applications (healthcare, finance, public sector)
- Auditing, compliance & regulation (EU AI Act, GDPR)
- Responsible AI governance and risk

SUBMISSION GUIDELINES

The workshop welcomes the following types of contributions:

- Regular research papers (up to 10 pages, including references), presenting mature and validated research contributions;

- Short research papers or position papers (up to 4 pages, including references), describing emerging ideas, ongoing research, preliminary results, or exploratory work.

Each submission should follow the CIKM 2026 workshop formatting guidelines.

ORGANIZERS

* Gabriele Tolomei, Sapienza University of Rome, Italy
* Ziheng Chen, Walmart Global Tech, USA
* Flavio Giorgi, tellmewAI, Italy
* Vittoria Vineis, Sapienza University of Rome, Italy
* Matteo Silvestri, Sapienza University of Rome, Italy
* Fabiano Veglianti, Sapienza University of Rome, Italy
* Lorenzo Antonelli, Sapienza University of Rome, Italy
* Edoardo Gabrielli, Sapienza University of Rome, Italy

Researchers interested in joining the Technical Program Committee are welcome to contact the workshop organizers.